Ridiculous dice

[thewarlordess]

One easy way to put it to rest is showing us the code. Is it open source anywhere?

It's not fully open source, but not closed source either. Volunteer developers can get access to the source.

There is a comment in the BGA FAQ: https://boardgamearena.com/faq?anchor=f ... ing_random

However, pointing out these facts is not enough to convince some people - if someone has convinced themselves that they've "seen" a non-random sequence of events, what evidence will convince them that your perception is wrong? Particularly if their understanding of probability, statistics, randomness and psychology/neuroscience is far lower than they think, and they don't know that the brain is wired to recognise patterns, even when there is objectively no pattern (e.g. white noise).

"Some people" meaning me? I have been playing long enough to now be able to predict what dice will come a lot of the times and for the other I have been making notes of number of doubles each player gets and it is definetly not random. We are not talking a few more for 1 player, more like 17-4 (in 3 point game) Its not normal for the other player to get 5 doubles in a row at the end eventhough he played shit throughout. Its not perception, it can be easily be confirmed by checking past games

[Neep]

I can't pull the individual game stats, but broadly speaking, 5 doubles in a roll is a .013% chance. My average number of rolls per game is 45.4; call it 45. The odds of 5 doubles somewhere in 45 rolls is roughly 1 in 227. I've played 4262 games, which means I have probably seen 5 doubles in a row 18-19 times (18.78). Divide the number of games played on BGA (7,430,826) by 227 and you can estimate that it has probably happened 32,735 times.

You can get a detailed explanation with your favorite AI. I used the prompt "What are the odds of rolling doubles 5 times in a row over 45 rolls? Show your work." in ChatGPT, and spot checked the numbers/logic.

[Romain672]

By the way, now that you have made claims, you can collect stats from now on, and share it after that.
That will remove most biais to only take future games.

And if there is really a problem (which is for us unlikely since so many people are bad at stats in Backgammon's forum), you should see something and you can share it.
If your stats with our calculation have a less than 1% chance, at least myself will investigate more (if you want to do your calculation, just do a clean claim now about what exactly is wrong and what you will be tracking).

[Jontia]

Do you have any proof that the dice are "unfair"?

Well for one I have been playing long enough to now be able to predict what dice will come a lot of the times and for the other I have been making notes of number of doubles each player gets and it is definetly not random. Its not normal for the other player to get 5 doubles in a row at the end eventhough he played shit throughout

Firstly, dice rolls are independent of play quality. So every player has the same chance of doubles no matter how shit their play throughout the game.

Secondly the chances of 5 doubles in a row are around 1 in 8,000. There have been 7.5 million games of backgammon on BGA since the game launched in 2014, which is just under 2,000 games a day. So on BGA as a whole, someone should get five doubles in a row twice a week.

Unusual things happen. That's what random means. If only expected things happened, that wouldn't be random.

If you're interested, you can take the rolls from a game and stick them into dice checkers available on the web. Like this one; https://dicefairness.johnellmore.com/ let us know how you get on.

[euklid314]

Do you have any proof that the dice are "unfair"?

Well for one I have been playing long enough to now be able to predict what dice will come a lot of the times and for the other I have been making notes of number of doubles each player gets and it is definetly not random. Its not normal for the other player to get 5 doubles in a row at the end eventhough he played shit throughout

Firstly, dice rolls are indpendently of play quality. So every player has the same chance of doubles no matter how shit their play throughout the game.

Secondly the chances of 5 doubles in a row are around 1 in 8,000. There have been 7.5 million games of backgammon on BGA since the game launched in 2014, which is just under 2,000 games a day. So on BGA as a whole, someone should get five doubles in a row twice a week.

Unusual things happen. That's what random means. If only expected things happened, that wouldn't be random.

If you're interested, you can take the rolls from a game and stick them into dice checkers available on the web. Like this one; https://dicefairness.johnellmore.com/ let us know how you get on.

Your calculation is based on BG games consisting of only 5 rolls (and having only one player).

Since there are many 5-long streaks in a single game (I calculated with 20 such streaks for each player, i.e. game lenght of 48 moves) the occurrences of 5 consecutive doubles are actually much higher.

One expects approx. 1 out of 250 games to have a streak of 5 doubles for one of the players (the expected value is 0.00514, i.e., 1 out of 194, but there can be several in a single game).

Thus there are approx. 8 such occurrences a day on BGA. And every player with hundreds of games played has seen such sequences. I definitely have seen it several times in my real life tournament play with (physical) precision dice.

[Jontia]

Your calculation is based on BG games consisting of only 5 rolls (and having only one player).

Thus there are approx. 8 such occurrences a day on BGA. And every player with hundreds of games played has seen such sequences. I definitely have seen it several times in my real life tournament play with (physical) precision dice.

All good points. If I wanted to ass cover, I might explain my simplistic approach by saying this is the chance of 'ending' on five doubles, so just the last five rolls for one player. But that would be a complete fiction made up to excuse my underestimates of the liklihood.

[euklid314]

Your calculation is based on BG games consisting of only 5 rolls (and having only one player).

Thus there are approx. 8 such occurrences a day on BGA. And every player with hundreds of games played has seen such sequences. I definitely have seen it several times in my real life tournament play with (physical) precision dice.

All good points. If I wanted to ass cover, I might explain my simplistic approach by saying this is the chance of 'ending' on five doubles, so just the last five rolls for one player. But that be a complete fiction made up to excuse my underestimates of the liklihood.

Touche! Well answered. The original discussion was about five doubles at the end of the game (it was not clear of OP if he meant exactly the last five rolls that finished the game)...

But then you have to take into consideration that doubles cluster at the end of a game!

I am dead serious that everybody of us has more doubles at the end of a game (say within the last five moves of a game)! Of course not only on BGA but in any live game as well. The teacher in me must leave it as a "homework" to the interested reader to figure out why.

My guess is that the probability of getting a double roll in the last five moves of a game is well over 20% (and not 16.67% as the unsuspicious player might expect)!

[MattTheHoser]

Every one of these threads ends up with people pointing out the stats on the dice rolls. What is left out of the frustration-discussion is the timeliness of the dice rolls in comparison to what was forced on the board. There are scenarios that will force you to leave a solo when you'd rather not. I'd love to see the stats on how often that gets punished with the perfect two dice scenario, navigating blank spaces on the board. There are far too many instances of a perfect posterior extraction roll that is punishing to the result of the game. Sure, one of two dice hitting perfect is expected, but the sniper shot of a two dice roll seems very unbalanced. There seems to be some dreadful if/then logic behind the randomness.

<flame away>

[Paddles]

"Some people" meaning me? I have been playing long enough to now be able to predict what dice will come a lot of the times and for the other I have been making notes of number of doubles each player gets and it is definetly not random.

The comment wasn't specifically aimed at you, because I don't know whether you are open to accepting mathematical evidence or not.

Others have already responded well to the comments about doubles, I won't duplicate them.

As to your other comment about being able to predict what dice will come. I'm happy to validate your claim by playing a sequence of games against you. You can type your predictions about the dice into the chat before each roll, then we'll see whether your predictions are accurate at a statistically significant level (i.e. your predictions should be correct too often to be simply chance; I could predict "you'll roll at least one six" and I'd be correct 3 times in 10 if the dice were totally fair). What times of day would be suitable (and in what timezone?). If we aren't able to be online at the same time, then hopefully someone else in this thread can make a similar offer.

The only condition is that we agree to publish the results no matter how they come out, and provide the raw data so others can validate or refute any conclusions.

[BarnardsStar]

There seems to be some dreadful if/then logic behind the randomness.

You can get a BGA Studio developer account and look at the source code for most (if not all) of the games. You’ll see there is no such ”if/then logic behind the randomness”. The game code asks the framework for two random numbers 1-6 inclusive. The random number function doesn’t know who’s asking. It doesn’t even know if this is for Backgammon or some other game that uses 2d6. It certainly doesn’t know the state of the board.

So if there is any such “logic” it would have to be in the game code, and it simply is not. No such thing, not there. You can look at the code and see.